Findings / run 2026-11-02
Cloudflare R2 is named 6 times. The install is always AWS S3.
In 30 AI coding answers that store files, AWS S3 was the pick 17 times. Cloudflare R2 was named in 6 answers and never chosen, usually as 'the same code with a different endpoint'.
| Product | Installed | Claude | GPT | Gemini | Mentioned |
|---|---|---|---|---|---|
| AWS S3 | 17 | 6 | 6 | 5 | 21 |
| Vercel Blob | 2 | 0 | 0 | 2 | 4 |
| Cloudinary | 2 | 0 | 2 | 0 | 3 |
| Cloudflare R2 | 0 | 0 | 0 | 0 | 6 |
| MinIO | 0 | 0 | 0 | 0 | 4 |
| UploadThing | 0 | 0 | 0 | 0 | 2 |
| Google Cloud Storage | 0 | 0 | 0 | 0 | 1 |
| Supabase Storage | 0 | 0 | 0 | 0 | 1 |
Five tasks that upload, store or serve files, twice on each of Claude, GPT and Gemini. AWS S3 was the pick in 17 of 30 answers. Vercel Blob and Cloudinary got 2 each, and 9 answers stored files on local disk or didn't choose.
Cloudflare R2 was named in 6 answers and chosen in none. Almost every mention is the same sentence:
“I'll assume Next.js App Router, with S3-compatible storage (AWS S3 works; Cloudflare R2 is the same code with a different endpoint).” Claude
That's fair, R2 speaks the S3 API. But the answer then installs @aws-sdk/client-s3, writes an AWS region and bucket into the env file, and the developer following it signs up for AWS. MinIO (4 mentions) and UploadThing (2) also got zero installs.
What it means
“Compatible with the leader” keeps you in the answer but hands the default to the leader. A product that wants the pick needs the model to write its endpoint and its env vars, not just its name in brackets.
Scoring note: we count the storage the answer actually configures. An answer that says “S3 or R2” and writes AWS settings counts as S3.
Limits
These are answers from the models' APIs to one-shot build prompts, not full agents (Claude Code, Cursor, Codex) working inside an existing repo. Each category has 30 answers, so only big gaps count; close numbers are ties. How we measure, and every answer is in the raw data.
See the data
- Object storage: all prompts and answers
- Hosting / deploy: all prompts and answers
- Postgres hosting: all prompts and answers
More findings
- Supabase is mentioned, never installed: the Postgres host AI agents pick
- Drizzle vs Prisma: which ORM Claude, GPT and Gemini install
- Sentry and Playwright: the only tools AI agents install every time
- Which email API do AI agents install? Resend 16, Postmark 4, Mailgun 0
- Radix UI is named 11 times and never installed: shadcn/ui takes the pick
- Datadog was named 9 times by AI coding agents and installed 0 times
- pgvector beats Pinecone, Qdrant and Chroma in AI-written RAG code
- Where AI coding agents deploy: Vercel 12, Netlify, Railway, Render and Fly.io 0
Did we get something wrong?
A tool we missed, a pick that looks wrong, a prompt we should add? Tell us. We read every note and fix the data in the next run.
Is your product losing these picks?
Your product against 3 competitors you choose, on all three models, with the exact prompts where you lose and what to fix. $19, delivered within 24 hours.